A surface plasmon polariton detector is demonstrated at infra-red wavelengths. The device consists of a metal stripe on silicon forming a Schottky contact thereon and supporting surface a plasmon polariton mode that is strongly confined and localised to the metal–semiconductor interface. Detection of optical radiation below the bandgap of silicon (at infrared wavelengths) occurs through internal photoemission. Responsivities of 0.38 and 1.04 mA/W were measured via end-fire coupling to a tapered optical fibre, at room temperature and at a wavelength of 1280 nm, for gold and aluminium stripes on n-type silicon, respectively. The device can be integrated with other structures used in nano-plasmonics, nano-photonics or silicon-based photonics, and it holds promise for short-reach optical interconnects and power monitoring applications.
© 2010 OSA
Over recent years, the field of plasmonics has undergone intensive research efforts . Plasmonics consists of the study of the properties and applications of surface plasmon polaritons (SPPs), which are optical surface waves propagating along the interface between a metal and a dielectric. The SPP is a coupled excitation propagating as a charge density wave in the metal coupled to electromagnetic fields, with its fields decaying evanescently in both the metal and the dielectric . The SPP has many interesting properties, including field confinement on a sub-wavelength scale  and very high surface and bulk sensitivities . These properties have resulted in many applications of SPPs in waveguiding [4–10], integrated optics [11–14], and sensing , to name a few. Developments on waveguiding of SPPs have shown, for instance, that a thin metal stripe of finite width embedded in a homogeneous dielectric cladding (symmetric structure) can support a long-range SPP (LRSPP), which propagates with lower loss and confinement than the single interface SPP [4,16]. A metal stripe with different dielectric claddings above and below the stripe (asymmetric structure) will also support bound SPP modes, but with higher confinement and loss than the LRSPP [5–7,9,10].
Selecting a lightly doped semiconductor as (one of) the dielectric cladding(s) of the metal stripe allows the formation of a Schottky contact at the material interface , conferring additional functionality to the structure. The Schottky diode structure has been widely investigated as an infra-red detector, for photon energies below the bandgap of the semiconductor . Numerous enhancements to the basic structure have been proposed, through resonant cavity enhancement [18,19], by SPP excitation [20–22], and by merging with a dielectric waveguide . Integration of a Schottky barrier detector with a metal stripe waveguide supporting SPPs has been investigated theoretically for symmetric  and asymmetric  silicon claddings. In this paper we demonstrate the operation of the asymmetric structure, and we present and discuss experimental results quantifying its performance as a SPP detector at photon energies below the bandgap of Si.
2. Device structure and operation
Figure 1(a) shows an isometric sketch of the device, with the metal stripe of width w, length l and thickness t acting simultaneously as a SPP waveguide and a Schottky contact to the underlying n-type Si (n-Si). The detection mechanism is internal photoemission: Absorption of SPPs by the metal stripe results in conduction electrons therein becoming excited (hot electrons), and if the energy hν (h - Planck’s constant, ν - optical frequency) of the SPP is sufficiently large, the electrons can gain sufficient energy to cross the Schottky barrier ФB, and be emitted into the semiconductor where they are collected as photocurrent [24,25]. Thus detection occurs for photon energies below the bandgap of the semiconductor but larger than the Schottky barrier height. (If the energy is greater than the bandgap, then electron-hole pair creation in the semiconductor dominates the photoresponse.)
The structure is designed to operate in the asb0 mode [5,25], sketched on the face of the structure in Fig. 1(b). This mode is tightly confined and strongly localised to the metal-semiconductor interface. As it propagates, it produces hot electrons in the metal right along the interface, thus enhancing the emission probability over the Schottky barrier.
As sketched in Fig. 1(b), the asb0 mode is excited by end-fire coupling with a PM (Polarization-Maintaining) tapered optical fibre. The fibre selected for the experiments has a ~2.5 µm spot size at a free-space optical wavelength of λ0 = 1550 nm . The coupling efficiency γc of this arrangement is estimated by calculating the overlap integral between the fields Fig. 1(b)), and ∞ denotes integration over the entire computational domain. The field distribution of the asb0 mode is obtained using a commercially available finite element mode solver.
Adopting Au and Al as the metals for the stripe, the asb0 mode fields were computed over the wavelength range from 1280 to 1620 nm, interpolating (cubic spline) the optical parameters from refractive index data available at several wavelengths . Figure 2(a) shows the Ey field distribution of the asb0 mode for Au and Al stripes of t ~135 nm and w = 2.5 µm; these dimensions should produce reasonable coupling efficiencies to the tapered fibre selected. The electric field is dominant in the vertical direction, so the mode is quasi-TM (Transverse Magnetic) in nature, as expected . For both metals, the fields are confined within the first micron below the stripe, with the confinement decreasing with increasing λ0.
The tapered fibre’s spot size also increases with wavelength, being ~10% larger at λ0 = 1550 nm than at 1310 nm . Interpolating the spot size (linearly), we compute the coupling efficiency to the asb0 mode over the λ0 range of interest assuming that the fibre is aligned with the bottom center of the metal stripe. As shown in Fig. 2(b), the coupling efficiency increases with wavelength due to the fields extending over a larger area and providing better overlap with the tapered fibre mode. Although the structure supports many higher order SPP modes  only the fundamental asb0 mode was found to be excited with relatively high coupling efficiency (up to ~35% for Al at λ0 = 1620 nm).
The computed mode power attenuation α of the asb0 mode ranges from α = 1.362 to 0.505 dB/μm, and α = 1.204 to 0.490 dB/μm, as λ0 goes from 1280 to 1620 nm, for Au and Al respectively. The high attenuation is due to the strong confinement of the mode, and ensures that a short waveguide will absorb essentially all of the in-coupled optical power.24]:19]):Eq. (5) versus hν is linear with an x-axis intercept of ΦB and a slope from which γc can be deduced. The condition αl >> 1 corresponds to the case where essentially all of the optical power coupled into the detector is absorbed.
The reverse bias leakage current (dark current) of the diode is 29]), and Carea is the Schottky contact area.
3. Fabrication and experimental set-up
The photodetectors were fabricated on a Si wafer consisting of a thin (15 μm) lightly n-doped layer (15 Ω-cm) epitaxially grown on a heavily n-doped wafer (0.01 Ω-cm). To create a bottom ohmic contact, a ~0.7 µm thick layer of Al was deposited on the backside, followed by a high temperature anneal. The metal stripes and features were then defined on the wafer’s topside using a UV bi-layer liftoff lithography technique . A mask aligner was used to expose the desired pattern on the wafer covered with layers of liftoff resist and photoresist. Immediately prior to top metal deposition, the wafers were placed in a buffered hydrofluoric acid solution to remove any native SiO2 layer and allow intimate contact between the metal and semiconductor. A thin metal film of t ~135 nm was deposited in an e-beam evaporation chamber under high vacuum (< 10−6 torr). The wafer was singulated into individual dies through careful cleaving. Figure 3(a) shows a SEM (scanning electron microscope) image of a die with an Al stripe, revealing a flat and smooth end facet suitable for optical coupling. An AFM (atomic force microscope) scan of an Al stripe, shown in Fig. 3(b), verified that the stripe had nearly uniform dimensions t = 135 nm and w = 2.5 µm; its roughness was measured as ~1.5 nm (root-mean-squared). The AFM results for an Au stripe were similar, except that the roughness was larger. Before testing the devices, the dies were cleaned in acetone and isopropyl alcohol, dried with nitrogen, and attached to a conductive carrier plane with conductive epoxy (Fig. 1(a)).
Testing was conducted by placing the die under microscope and using tungsten probes attached to micropositioners to make electrical contact to the device. The bias voltage is applied by probing a large probe pad on top of the device and the ohmic contact on its bottom surface (Fig. 1(a)). Another micropositioner was then used to align the tapered PM fibre (polarisation-aligned to launch TM light) to the waveguide, as per the arrangement shown in Fig. 1. The tapered fibre output field behaves as a focused Gaussian beam, reaching its narrowest spot size at the focal point which is about 14 μm away from the tip. The fibre was aligned to the waveguide by placing the device under reverse bias and maximising the measured photocurrent. As additional evidence of the excitation of the asb0 mode, it was observed that nearly no photocurrent was generated when the fibre was rotated to emit TE-polarised (Transverse-Electric) light.
Once the best possible alignment was achieved, a computer-automated routine controlling ancillary instrumentation was run to rapidly execute test routines and gather measurement data. The ancillary instrumentation consisted of two linearly polarised lasers operating near 1310 and 1550 nm each tunable over ~100 nm, a 2 × 2 fibre coupler with a 50:50 splitting ratio at λ0 = 1550 nm, an optical power meter, and a multimeter. The two lasers were connected to the coupler’s input ports while the output ports were connected to the tapered fibre and the power meter. With the coupler’s splitting ratio characterised across the entire λ0 range, the power meter was used to maintain a constant optical power incident onto the photodetector end facet (Pinc) as each laser wavelength was scanned. The multimeter served the dual purpose of applying a voltage bias and measuring the current. All measurements were obtained at room temperature.
4. Experimental results
The electrical performance of several fabricated detectors was determined by measuring the diode current (I) as a function of the applied voltage (V); Fig. 4 shows typical measurements. The detectors show the expected rectifying behaviour, with the slope in the forward bias region limited by the detectors’ series resistance (this resistance can be lowered by eg, placing ohmic ground contacts along the top surface near the Schottky contact instead of along the bottom). The turn-on voltage for the Au detector is about 0.3 V whereas that of the Al is near 0 V.
The inset of Fig. 4 reveals dark currents of 10 nA and 6 µA, for Au and Al, respectively. These dark currents are rather large because conduction occurs over the entire Schottky contact area, which includes the large probe pad along the top surface (Fig. 1(a)). The contact areas (including the pad) are Carea = 2.71 × 10−5 and 1.0085 × 10−4 cm2 for the Au and Al detectors, respectively. Isolation of the pads from the Si surface can be achieved by adding a thin intervening dielectric, which would result in a large reduction of the dark current because the pad occupies over 90% of the total contact area.
Solving Eq. (6) to obtain ФB is not very accurate because the experimental setup and the detectors have a high resistance. There are several techniques for extrapolating the barrier height of high resistance diodes, based on forward bias measurements . The methods proposed by Lien et al. , and Werner  have been shown to yield accurate results. Following the Lien et al.  method yields ФB = 0.7573 and 0.4447 eV for the Au and Al detectors, respectively, whereas the Werner method yields ФB = 0.7444 and 0.4809 eV, respectively. The ideality factors  were found to be 1.33 and 1.08, respectively. These ФB’s differ from the expected values (0.8 and 0.72 for Au and Al respectively ); however ФB is known to be very sensitive to the method and conditions of contact formation . Given that hν > ФB must be satisfied in order for the SPP to be detected, the long-wavelength cut-off of the detectors are λ0 ~1650 and 2700 nm, respectively.
The detectors were then characterised optically, first by measuring the diode current I versus Pinc under reverse bias at several λ0. Figure 5 shows the measurements obtained at a bias of V = −100 mV for the same detectors used in Fig. 4. We emphasise that nearly no photocurrent was generated when the PM tapered fibre was rotated to emit TE-polarised light, so we attribute the photocurrent to the absorption of the SPP mode (asb0) as it propagates along the waveguide.
From Fig. 5, it is noted that -I is linear with Pinc. Recall that under a weak reverse bias, -I ≅ Idark + Iph = Idark + RPinc, so the slope of these curves yields the responsivity; they are summarised in Table 1 (under Exp.). In the case of the Au detector, Iph >> Idark ~10 nA over most of the measurement range, and R is observed to decrease with increasing wavelength. The longest wavelength tested (λ0 = 1620 nm) is just above cut-off. In the case of the Al detector, Iph remains only slightly above Idark ~6 μA over most of the measurement range (due to power limitations in the lasers used in the set-up and the large dark current of this detector), and R is observed to depend less strongly on wavelength. The responsivity and dark current of the Al detector are larger than those of the Au detector because the Schottky barrier of the former is lower.
Maximum responsivities of 0.38 and 1.04 mA/W are measured for the Au and Al detectors, respectively, at the shortest wavelength, λ0 = 1280 nm (Table 1). These values are comparable to those reported in Ref . for a NiSi2 on p-Si Schottky barrier detector integrated into a Si on insulator waveguide (4.6 mA/W at λ0 ~1550 nm).
The wavelength response of the detectors was then measured at a constant Pinc = 2 mW and at a bias of V = −100 mV; Figs. 6(a) and 6(c) give the measurements. (The wavelength ranges investigated are dictated by the tunable lasers available.) The measurements of the Al detector (Fig. 6(c)) for λ0 ~1550 nm are noisier (and less reliable) than those of the Au detector (Fig. 6(a)), again because Idark is larger for the former.
Figures 6(b) and 6(d) plot the corresponding R1/2hν versus hν (such plots are termed Fowler plots - eg ). The measurements approximately follow linear models (the best fitting ones are shown), except for the Al detector at λ0 ~1550 nm (Fig. 6(d)) because Idark is large. As pointed out following Eq. (5), if γc is independent of ν, and αl >> 1 over the range of ν considered, then the plot is linear with an x-axis intercept of ΦB and a slope from which γc can be deduced. The condition αl >> 1 is essentially satisfied for the detectors characterised given their length (l = 80 and 35 µm for Au and Al) and attenuation (see Section 2); αl > 4 over the wavelength range of interest for both. However, γc depends on ν as shown by the computations of Fig. 2(b), explaining the observed slight deviation from linearity (especially in the better quality measurements of Fig. 6(b)).
The x-axis intercepts of the linear models on the Fowler plots yield ФB = 0.765 and 0.494 eV for the Au and Al detectors, respectively; these values are in good agreement with those obtained from I-V measurements. From the slope of the linear models, we find γc = 3.85% and 2.01% for the Au and Al detectors, respectively. These coupling efficiencies are lower than those estimated theoretically (Fig. 2(b)). Part of the discrepancy is attributable to alignment difficulties (the asb0 mode fields extend to <1 µm vertically - Fig. 2(a)), and part is attributable to a simplified theory (the coupling calculations neglect possible beam curvature at the detector end facet, and the expression for ηi, Eq. (3), neglects the attenuation length of hot carriers).
Finally, the measured responsivities are compared with theoretical ones computed via Eqs. (2) and (3) using experimental values for γc and ФB, and the computed values for α given in Section 2. The theoretical results are summarized in Table 1 (under Theo.). Considering that several values were obtained for ФB, and that γc was estimated from the slope of the Fowler plot, these parameters were adjusted slightly to provide the best agreement between the experimental and theoretical responsivities: ФB was taken as 0.744 and 0.484 eV for the Au and Al detectors respectively, and γc was taken as 4.3% and 2.5%, respectively (all are close to the measured values). Average errors of 33.9% and 8.8% are obtained for the Au on n-Si and Al on n-Si detectors, respectively. The agreement is better at the shorter wavelengths (λ0 ~1310 nm) than at the longer ones (λ0 ~1550 nm).
5. Summary and concluding remarks
In summary, SPP Schottky contact detectors were fabricated on n-Si, characterised, and compared to theoretical predictions. An asymmetric metal stripe waveguide was designed to operate in a strongly confined SPP mode localised at the metal - Si interface (the asb0 mode). The mode propagates with a high attenuation of 1.362 and 1.204 dB/μm for Au and Al on n-Si, respectively, so the detectors are very short in length, 80 and 35 µm, respectively, for complete absorption of the coupled optical power. Respective dark currents of 10 nA and 6 µA were measured, which can be reduced by isolating the device probe pads from the Schottky contact plane. Responsivities of 0.38 and 1.04 mA/W were measured for Au and Al on n-Si respectively, at λ0 = 1280 nm. These values are lower than theoretically possible, partly because the SPP mode was excited low coupling efficiency. The fabricated detectors have been successfully demonstrated across a significant portion of the infrared spectrum. The detectors can be integrated with other structures used in plasmonics or silicon-based photonics, and hold promise for short-reach optical interconnects and power monitoring applications.
The authors thank Rob Vandusen and Carol Adams for numerous comments and insight provided during fabrication. The input and help of Ewa Lisicka-Skrzek during the measurements is also gratefully acknowledged.
References and links
2. S. A. Maier, Plasmonics: Fundamentals and Applications (Springer, New York, USA, 2007).
3. P. Berini, “Bulk and surface sensitivities of surface plasmon waveguides,” N. J. Phys. 10(10), 105010 (2008). [CrossRef]
4. P. Berini, “Plasmon-polariton waves guided by thin lossy metal films of finite width: Bound modes of symmetric structures,” Phys. Rev. B 61(15), 10484–10503 (2000). [CrossRef]
5. P. Berini, “Plasmon-polariton waves guided by thin lossy metal films of finite width: Bound modes of asymmetric structures,” Phys. Rev. B 63(12), 125417 (2001). [CrossRef]
6. B. Lamprecht, J. R. Krenn, G. Schider, H. Ditlbacher, M. Salerno, N. Felidj, A. Leitner, F. R. Aussenegg, and J. C. Weeber, “Surface plasmon propagation in microscale metal stripes,” Appl. Phys. Lett. 79(1), 51–53 (2001). [CrossRef]
7. J.-C. Weeber, J. R. Krenn, A. Dereux, B. Lamprecht, Y. Lacroute, and J. P. Goudonnet, “Near-field observation of surface plasmon polariton propagation on thin metal stripes,” Phys. Rev. B 64(4), 045411 (2001). [CrossRef]
8. T. Nikolajsen, K. Leosson, I. Salakhutdinov, and S. I. Bozhevolnyi, “Polymer-based surface-plasmon-polariton stripe waveguides at telecommunication wavelengths,” Appl. Phys. Lett. 82(5), 668–670 (2003). [CrossRef]
9. R. Zia, M. D. Selker, and M. L. Brongersma, “Leaky and bound modes of surface plasmon waveguides,” Phys. Rev. B 71(16), 165431 (2005). [CrossRef]
11. R. Charbonneau, C. Scales, I. Breukelaar, S. Fafard, N. Lahoud, G. Mattiussi, and P. Berini, “Passive integrated optics elements based on long-ranging surface plasmon polaritons,” J. Lightwave Technol. 24(1), 477–494 (2006). [CrossRef]
12. B. Steinberger, A. Hohenau, H. Ditlbacher, F. R. Aussenegg, A. Leitner, and J. R. Krenn, “Dielectric stripes on gold as surface plasmon waveguides: Bends and directional couplers,” Appl. Phys. Lett. 91(8), 081111 (2007). [CrossRef]
13. T. Holmgaard, Z. Chen, S. I. Bozhevolnyi, L. Markey, A. Dereux, A. V. Krasavin, and A. V. Zayats, “Bend- and splitting loss of dielectric-loaded surface plasmon-polariton waveguides,” Opt. Express 16(18), 13585–13592 (2008). [CrossRef]
14. T. W. Ebbesen, C. Genet, and S. I. Bozhevolnyi, “Surface plasmon circuitry,” Phys. Today 61(5), 44–50 (2008). [CrossRef]
15. J. Homola, S. S. Yee, and G. Gauglitz, “Surface plasmon resonance sensors: review,” Sens. Actuactor. B 54(1-2), 3–15 (1999). [CrossRef]
16. P. Berini, “Long-range surface plasmon-polaritons,” Adv. Opt. Photonics 1(3), 484–588 (2009). [CrossRef]
17. S. M. Sze, and K. K. Ng, Physics of Semiconductor Devices (Wiley, New York, USA, 2006).
18. M. Casalino, L. Sirleto, L. Moretti, M. Gioffrè, G. Coppola, and I. Rendina, “Silicon resonant cavity enhanced photodetector based on the internal photoemission effect at 1.55 µm: Fabrication and characterization,” Appl. Phys. Lett. 92(25), 251104 (2008). [CrossRef]
19. H. Elabd and W. F. Kosonocky, “Theory and measurements of photoresponse for thin film Pd2Si and PtSi infrared Schottky-barrier detectors with optical cavity,” RCA Review 43, 569–589 (1982).
20. S. R. J. Brueck, V. Diadiuk, T. Jones, and W. Lenth, “Enhanced quantum efficiency internal photoemission detectors by grating coupling to surface plasma waves,” Appl. Phys. Lett. 46(10), 915–917 (1985). [CrossRef]
22. C. Daboo, M. J. Baird, H. P. Hughes, N. Apsley, and M. T. Emeny, “Improved surface plasmon enhanced photodetection at an Au-GaAs Schottky junction using a novel molecular beam epitaxy grown Otto coupling structure,” Thin Solid Films 201(1), 9–27 (1991). [CrossRef]
23. S. Zhu, M. B. Yu, G. Q. Lo, and D. L. Kwong, “Near-infrared waveguide-based nickel silicide Schottky-barrier photodetector for optical communications,” Appl. Phys. Lett. 92(8), 081103 (2008). [CrossRef]
25. A. Akbari and P. Berini, “Schottky contact surface-plasmon detector integrated with an asymmetric metal stripe waveguide,” Appl. Phys. Lett. 95(2), 021104 (2009). [CrossRef]
26. Oz Optics, Tapered PM Optical Fiber (TPMJ-X-1550–8/125–0.4–10–2.5–14–1) (www.ozoptics.com)
27. A. B. Buckman, Guided-Wave Photonics (Harcourt Brace Jovanovich, New York, USA, 1992).
28. E. D. Palik, ed., Handbook of Optical Constants of Solids (Academic, Orlando, USA, 1985).
29. P. Kramer and L. J. van Ruyven, “Position of the band edges of silicon under uniaxial stress,” Appl. Phys. Lett. 20(11), 420–422 (1972). [CrossRef]
30. P. Berini, N. Lahoud, and R. Charbonneau, “Fabrication of surface plasmon waveguides and integrated components on ultrathin freestanding membranes,” J. Vac. Sci. Technol. A 26(6), 1383–1391 (2008). [CrossRef]
31. V. Aubry and F. Meyer, “Schottky diodes with high series resistance: Limitations of forward I-V methods,” J. Appl. Phys. 76(12), 7973–7984 (1994). [CrossRef]
32. C.-D. Lien, F. C. T. So, and M.-A. Nicolet, “An improved forward I-V method for nonideal Schottky diodes with high series resistance,” IEEE Trans. Electron. Dev. 31(10), 1502–1503 (1984). [CrossRef]
33. J. H. Werner, “Schottky barrier and pn-junctionI/V plots - Small signal evaluation,” Appl. Phys., A Mater. Sci. Process. 47(3), 291–300 (1988). [CrossRef]
34. J. M. Mooney, “The dependence of the Schottky emission coefficient on reverse bias,” J. Appl. Phys. 65(7), 2869–2871 (1989). [CrossRef]